W3C home > Mailing lists > Public > public-media-annotation@w3.org > April 2011

OWL2 profiles, another proposed minor change

From: Pierre-Antoine Champin <pierre-antoine.champin@liris.cnrs.fr>
Date: Wed, 06 Apr 2011 11:28:52 +0200
Message-ID: <4D9C3254.20905@liris.cnrs.fr>
To: "public-media-annotation@w3.org" <public-media-annotation@w3.org>
Hi again,

while I am at it, I was wondering whether our RDF ontology is also
compliant with other profiles of OWL2.

Recall that we are considering for the moment OWL2-RL, which can be
implemented on top of a rule engine.

Another interesting profile is OWL2-QL, which can be implemented on top
of a relational database, which I think is a very nice feature.

The *only* thing that prevents us from being compliant with OWL2-QL (in
addition to OWL2-RL) is that we are using the datatype xsd:double
(supported in the latter, but not the former -- no Felix, don't ask
me...). However, OWL2-QL supports xsd:decimal (which is also supported
by other profiles).

Note that we don't lose semantics: the value space of xsd:decimal is a
superset as xsd:double, as xsd:decimal has no size limit. However, and
this is the bothering part, the lexical space of xsd:decimal does not
contain the lexical space of xsd:double, because xsd:decimal does not
support 'exponent' notation, like "1.2e-3", you have to write
"0.0012"(personal opinion: we can live with that!).

I see several options here:

a) we don't care about OWL2-QL, and leave things as they are.

b) we only replace xsd:double by xsd:decimal in the RDF only, and
document it, arguing that we don't lose semantics, and gain OWL2-QL
compliance ; but we keep the definitions of the properties as is. Easier
from an editorial point of view, but a bit awkward.

c) we replace all mentions to 'double' by 'decimal' in the ontology

Note that we don't have to change the API document: WebIDL has its own
datatypes, and double is the best match to either xsd:double or xsd:decimal

Received on Wednesday, 6 April 2011 09:29:17 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:17:41 UTC